Several fixes and enhancements

[musashiXXX]

The changes in this pull request fix:

• newTable needed to be renamed to new_table in ETLAlchemySource.py
• schema_transformer.py was not correctly retrieving the table name from the table_transformations dict, which would effectively disable all mappings in the table_schema_transformation_file.
• MSSQL UNIQUEIDENTIFIER types will now be converted to PostgreSQL UUID types when migrating from MSSQL -> PostgreSQL
• Added a per_table_buffers kwarg that allows for specifying the number of rows to fetch at once from the source database
• We now check the varchar_length and try to adjust it for our target database, e.g., MSSQL varchar(max) -> PostgreSQL text

[seanharr11]

@musashiXXX thank you for the PR! One concern:

1. I am not in agreement with your generalization of coercing data to a uuid.UUID if isinstance(bytearray, value) is True. While this likely solved your use case, I am concerned that it doesn't generalize to other columns that hold bytearray values. There may be a BINARY TEXT blob column in a DB, and we don't necessarily want to turn that into a UUID.

[musashiXXX]

@seanharr11

I am not in agreement with your generalization of coercing data to a uuid.UUID if isinstance(bytearray, value) is True.

I agree. And the 8th commit in that PR un-does this behavior. The PR with the code you mentioned was submitted in haste. See here. There was this commit containing the ill-conceived coerce all bytea to uuid code, and then two commits later the correction.